Island shallow sea topographic surveying and mapping method based on active-passive remote sensing fusion

Abstract

The invention relates to the technical field of ocean remote sensing monitoring, and particularly discloses an island shallow sea topographic surveying and mapping method based on active-passive remote sensing fusion, which comprises the following steps of: 1, acquiring active-passive data; 2, performing data preprocessing on the active data and the passive data respectively; 3, constructing an optical image data set; and 4, constructing a water depth inversion model based on the water depth data and the optical image data set. According to the method, the advantages of wide coverage of passive remote sensing and high precision of active remote sensing are fused, and the water depth information of shallow sea islands and reefs can be accurately obtained in a large-area range. In practical application, complex professional equipment does not need to be arranged, a large amount of manpower does not need to be input, the operation cost and difficulty are greatly reduced, depth sounding and surveying and mapping tasks of the shallow sea island reef area can be rapidly completed, and practicability is high. The problems in the prior art that shallow sea island monitoring is difficult, the cost is high, and the requirement for the monitoring environment is high are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean remote sensing monitoring, and in particular relates to a method for surveying and mapping island and shallow sea terrain based on active-passive remote sensing fusion. Background Art

[0002] High-precision depth measurements in offshore and island areas play a key role in many fields. They are of great significance to navigation safety. Accurate depth data can help ships avoid hazards such as reefs and ensure the smooth operation of maritime transportation. In terms of resource exploration, it helps to discover potential marine resources and promote the development of the marine economy. In terms of environmental protection, it can provide an important basis for marine ecological research and help maintain the ecological balance of the ocean. Monitoring islands and reefs can help us understand the impact of human and natural factors on islands and reefs. Natural factors, such as monsoons, typhoons, seawater erosion and sediment deposition, usually have long-term and slow impacts on islands and reefs; while human factors, such as land reclamation, overfishing, and resource development, can lead to more rapid and direct ecological changes. In addition, this type of monitoring is also important for revealing trends in environmental change and responding to marine events (such as storms and tsunamis).

[0003] Acoustic bathymetry is traditionally used to monitor reefs in coastal areas and islands. However, its application in these areas still faces numerous limitations. Specifically, it is expensive, requiring significant investment in equipment, maintenance, and labor; it is highly sensitive to sea conditions, making accurate measurements difficult in harsh seas; and it has stringent environmental requirements, making it difficult to meet measurement requirements in some complex waters. These issues make high-precision remote sensing of water depth both a challenging and challenging area in ocean exploration. Furthermore, traditional ocean bathymetry relies primarily on instruments based on single-beam and multi-beam techniques. This measurement method has limitations in environmental adaptability and operator comfort, making it difficult to meet the demands of modern ocean exploration. However, advances in remote sensing technology have brought new opportunities for depth inversion, particularly optical remote sensing, which can rapidly cover extensive ocean areas and collect large amounts of data, offering new possibilities for depth measurement.

[0004] Based on this, the present invention proposes a method for surveying and mapping shallow sea topography of islands and reefs based on active-passive remote sensing fusion to solve the problems of shallow sea island and reef monitoring difficulties, high costs, and high requirements for the monitoring environment in the above-mentioned existing technologies. Summary of the Invention

[0005] Technical problems solved

[0006] In response to the shortcomings of the existing technology, the present invention provides a method for mapping the shallow sea topography of islands and reefs based on the fusion of active and passive remote sensing. This method can accurately invert and map the water depth and topography of shallow islands and reefs; it solves the problems of shallow island and reef monitoring difficulties, high costs, and high requirements for the monitoring environment in the existing technology, and provides detailed and reliable basic geographic information data for many fields such as marine resource development, navigation safety, and coastal ecological protection.

[0007] Technical Solution

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for mapping island and reef shallow sea topography based on active and passive remote sensing, comprising:

[0010] Step 1: Get active-passive data

[0011] The active-passive data includes active data and passive data;

[0012] Step 2: Preprocess the active and passive data separately to extract effective water depth data;

[0013] Step 2.1: Crop and resample the passive data image, identify clouds, water bodies, and land in the image, and perform cloud and land masking;

[0014] Step 2.2: Perform DBSCAN photon clustering denoising, refraction correction, and tide correction on the active data to extract effective water depth data;

[0015] Step 3: Construct an optical image dataset;

[0016] Step 4: Based on the water depth data obtained in step 2 and the optical image dataset in step 3, a quadratic polynomial ratio model is used to construct a water depth inversion model.

[0017] In a preferred embodiment, the active data in step 1 is the Ice, Cloud and Land Elevation Satellite II ICESat-2 dataset; and the passive data is the Sentinel-2 dataset.

[0018] In a preferred embodiment, the process of cropping and resampling the passive data image, identifying clouds, water bodies, and land in the image, and performing cloud masking and land masking in step 2.1 includes:

[0019] Step 2.1.1: Crop and resample the satellite imagery data of the downloaded passive data in SNAP;

[0020] Step 2.1.2: After cropping and resampling the data, use the remote sensing image processing platform to calculate the normalized difference vegetation index. Set thresholds to identify clouds and land. Combined with visual interpretation, create cloud and land masks. Use regions of interest to manually annotate small cloud shadows to form a new mask layer and apply it to the image.

[0021] In a preferred embodiment, the process of performing DBSCAN photon clustering denoising, refraction correction and tidal correction on the active data in step 2.2 includes:

[0022] Step 2.2.1: Based on the confidence parameter in the photon point cloud file of the active data, remove the photons with a parameter of 0 to complete the coarse denoising;

[0023] Step 2.2.2: After completing the coarse denoising, use the Gaussian distribution model to perform fine denoising to separate the water surface and bottom photons;

[0024] Step 2.2.3: Use the density-based clustering algorithm DBSCAN to perform cluster analysis on the photon point group and extract the photon point groups on the sea surface and seabed.

[0025] In a preferred embodiment, the process of performing DBSCAN photon clustering denoising, refraction correction and tidal correction on the active data in step 2.2 further includes: step 2.2.4: establishing a data set for the bathymetric mapping task, including:

[0026] (1) For some DBSCAN algorithm extraction errors, human-computer interaction is used to correct the extraction errors;

[0027] (2) After obtaining the photon point group data, the improved Parrish water refraction correction geometric model is used to perform refraction correction to obtain the water depth measurement results. The tide correction model and tide data from the tide website are then used to perform tide correction on the bathymetry results and unify them to the mean sea level.

[0028] (3) Finally, the active data bathymetry results are sorted in ascending order, and one point is extracted every 10 data points as the data set for the subsequent bathymetry and mapping tasks of islands and reefs.

[0029] In a preferred embodiment, the process of constructing the optical image dataset in step 3 includes:

[0030] Step 3.1: Construct an image collection of passive data in time series and calculate the median logarithmic ratio of the blue-green bands of all images in each year. The formula is:

[0031]

[0032] Among them, Rblue and R green are the surface reflectances of the blue and green bands, respectively, and n is a constant;

[0033] Step 3.2: After obtaining the median of the blue-green band logarithmic ratio of all images, use the median calculation function to obtain the median matrix of the blue-green band logarithmic ratio of each image in each year to form a high-quality optical remote sensing image dataset.

[0034] In a preferred embodiment, the process of constructing the water depth inversion model in step 4 includes:

[0035] Step 4.1: Construct a quadratic polynomial ratio model based on the logarithmic ratio of the blue and green bands. The formula is:

[0036] H=a×W md 2 +b×W md +c

[0037] Among them, W md is the median of the logarithmic ratio of the blue and green bands calculated from passive data images within a certain time series; H represents the measured water depth data; a, b, and c represent the model fitting coefficients calculated by the least squares method;

[0038] Step 4.2: Use the nearest neighbor method to match the passive data and active data according to longitude and latitude, and divide the data into training data set and validation data set;

[0039] Step 4.3: Finally, the quadratic polynomial ratio model is trained using the training dataset data, and the model fitting results are verified and evaluated using the validation dataset data to obtain the water depth inversion model.

[0040] In a preferred embodiment, when dividing the training data set and the validation data set, a random sampling method is used to divide them according to a ratio of 7:3.

[0041] In a preferred embodiment, the method for mapping shallow sea topography of islands and reefs based on active-passive remote sensing further includes: Step 5: using the water depth inversion model trained in Step 4 to invert the water depth data of shallow sea islands and reefs.

[0042] In a preferred embodiment, the process of inverting the water depth data of shallow islands and reefs in step 5 includes:

[0043] Step 5.1: Input the passive data processed in step 2 into the water depth inversion model trained in step 4. Combined with the model parameters trained in step 4, the water depth data at different locations of the shallow islands and reefs are calculated to generate an accurate bathymetric map.

[0044] Step 5.2: Then, combine with the GIS system to mark the terrain features of the islands and reefs.

[0045] Beneficial effects

[0046] Compared with the existing technology, the present invention provides a method for mapping island and reef shallow sea topography based on active-passive remote sensing fusion, which has the following beneficial effects:

[0047] 1. This method, based on a fusion of active and passive remote sensing bathymetry and mapping, constructs a depth inversion model that has demonstrated excellent performance in validation results at shallow islands and reefs. In validation at Baijiao and Beizi Island, the coefficient of determination exceeded 0.9, the mean absolute error was less than 0.41m, and the root mean square error was less than 0.73m, meeting the requirements for high-precision depth measurements.

[0048] 2. The present invention is based on an active-passive remote sensing fusion bathymetric and mapping method, which cleverly combines the advantages of wide coverage of passive remote sensing and high precision of active remote sensing, and can accurately obtain water depth information of shallow islands and reefs over a large area. In practical applications, there is no need to equip complex professional equipment or invest a lot of manpower, which greatly reduces the cost and difficulty of operation, and can quickly complete the bathymetric and mapping tasks of shallow islands and reefs, with strong practicality. In particular, it provides important technical support for monitoring shallow sea topography, and effectively guarantees the efficient and continuous development of shallow island and reef bathymetric and mapping work. It solves the problems of shallow island and reef monitoring difficulties, high costs, and high requirements for the monitoring environment in existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a method for mapping island and shallow sea terrain based on active-passive remote sensing fusion implemented by the present invention;

[0050] Figure 2 This is a schematic diagram of the Sentinel-2 data processing process;

[0051] Figure 3 This is the result of ICESat-2 data processing;

[0052] Figure 4 This is the accuracy verification result of the water depth inversion model of Baijiao and Beizi Island;

[0053] Figure 5 This is the water depth inversion result of active-passive fusion data in Baijiao in 2018;

[0054] Figure 6 This is the water depth inversion result of active-passive fusion data in Beizi Island in 2018;

[0055] Figure 7 This is the topographic survey result of Beizi Island;

[0056] Figure 8This is the result of the Baijiao topographic survey.

[0057] in:

[0058] exist Figure 2 Middle: Figure (a) shows the preprocessed image collection; Figure (b) shows the calculation of the median blue-green band ratio; Figure (c) shows the final combined image.

[0059] exist Figure 3 Middle: Blue represents sea surface photon points, red represents seabed photon points, and gray represents noise photon points. The figure only shows the part of the strip containing the bathymetric results;

[0060] exist Figure 4 Medium: The model fit is good, with a higher coefficient of determination (R 2 The close to 1) and low MAE and RMSE jointly verify the accuracy and reliability of the model. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0062] Example 1:

[0063] See also Figures 1-8 This embodiment provides a technical solution: a method for mapping island and reef shallow sea terrain based on active-passive remote sensing fusion, comprising:

[0064] Step 1: Get active-passive data

[0065] In the active-passive data model, the active data is the Ice, Cloud, and Land Elevation Satellite II ICESat-2 dataset (from the Copernicus Data Center). This example uses the ICESat-2ATL03 dataset, which is Level 2. The passive data is the Sentinel-2 dataset (from NASA Earthdata), which is Level 2A. This example uses images with cloud coverage below 30% and minimal solar flares.

[0066] Step 2: Preprocess active and passive data separately to extract effective water depth data

[0067] Step 2.1: Crop and resample the passive data Sentinel-2 dataset image, use NDVI to identify clouds, water bodies and land in the image, and perform cloud masking and land masking. The specific process is as follows: Figure 2 As shown;

[0068] Step 2.1.1: Crop and resample the satellite image data of the downloaded passive data Sentinel-2 dataset in the Sentinels Application Platform (SNAP) developed by the European Space Agency (a cross-Linux distribution package management technology). First, open and load the downloaded image in SNAP, select "Raster"->"Subset..." in the toolbar to open the cropping tool, select the study area and the bands to be retained, and crop the image range and bands. Then select "Raster"->"Geometric Operation"->"Resampling" to open the resampling tool, select the cropped image as input, set the target resolution to 10m, and use the default nearest neighbor method to resample the image;

[0069] Step 2.1.2: After cropping and resampling the data, use Band Math in the Environment for Visualizing Images (ENVI) remote sensing image processing platform to calculate the Normalized Difference Vegetation Index (NDVI). Set thresholds to identify clouds and land, and combine visual interpretation to create cloud and land masks. Use regions of interest to manually annotate small cloud shadows, creating a new mask layer and applying it to the image.

[0070] Among them, in the above-mentioned remote sensing image processing platform ENVI (The Environment for Visualizing Images), the normalized difference vegetation index (NDVI) is used to distinguish water bodies and land, as well as cloud detection, and then manual masking is performed through visual interpretation to ensure that all clouds and cloud shadow areas are masked.

[0071] Step 2.2: The process of preprocessing the active data ICESat-2 dataset includes: selecting appropriate DBSCAN parameters, applying the DBSCAN algorithm to perform cluster analysis, identifying and removing noise points; performing refraction correction based on the atmospheric refraction model; performing tidal correction based on the tidal data at the time of image acquisition on the tidal website, and extracting effective water depth data. Specifically, it includes:

[0072] Step 2.2.1: Based on the confidence parameter in the photon point cloud file of the ICESat-2 dataset, remove the photons with a parameter of 0 to complete the coarse denoising.

[0073] Step 2.2.2: After completing the coarse denoising, fine denoising is performed using the Gaussian distribution model. First, the elevation distribution histogram of the photon point cloud data needs to be obtained. Based on the elevation distribution of the original photon point cloud data, a height distribution histogram can be obtained. Through this histogram, two Gaussian peaks can be identified. These two peaks correspond to the intervals where water surface photons and water bottom photons are located. Based on these two intervals, the water surface and water bottom photons are separated.

[0074] Step 2.2.3: Use the density-based clustering algorithm DBSCAN to perform cluster analysis on the photon point group and extract the photon point groups on the sea surface and seabed. Specifically include:

[0075] Two important parameters of the DBSCAN algorithm are Eps (neighborhood radius) and MinPts (minimum number of neighborhood points). MinPts determines the minimum number of points required for an area to be considered a cluster in the DBSCAN algorithm. Its setting is crucial for distinguishing signal photons from noise photons. The calculation formula is as follows:

[0076]

[0077] Here, SN1 represents the expected total photon number, determined by counting all photons within a given radius; SN2 represents the expected noise photon number, determined by counting photons at a minimum vertical elevation of 5 meters. This formula ensures that the algorithm can effectively distinguish between signal photons and noise photons under varying signal-to-noise ratios.

[0078] The neighborhood radius Eps value is usually determined by the K-average nearest neighbor method. This method sets a window containing 1000 points, then moves the window along the data trajectory, calculates the Euclidean distance between points in each window, and thus generates a distance matrix. The average value of the distance matrix is then taken. These average distances are the candidate Eps values.

[0079] Step 2.2.4: Create a data set for the bathymetric mapping mission

[0080] For some erroneous results extracted by the DBSCAN algorithm, human-computer interaction is used to correct the erroneous points, and a graphical user interface tool is used to visually identify and correct the erroneously extracted points, thereby obtaining data with higher accuracy. After obtaining the photon point group data, the improved Parrish water refraction correction geometric model is used to perform refraction correction to obtain the water depth measurement results. The tide correction model and the tide data on the tide website are then used to perform tide correction on the sounding results and unify them to the mean sea level. Finally, representative data points are selected from a large amount of sounding data, and the ICESat-2 sounding results are sorted in ascending order. One point is extracted every 10 data points as the data set for the subsequent sounding and mapping mission of the islands and reefs. In this way, the amount of data can be effectively reduced while retaining key information. The processing results are as follows: Figure 3 shown.

[0081] Step 3: Construct an optical image dataset, including:

[0082] Step 3.1: Construct an image collection of Sentinel-2 data in time series and calculate the median logarithmic ratio of the blue-green bands of all images in each year. The formula is:

[0083]

[0084] Among them, R blue and R green are the surface reflectances of the blue and green bands, respectively, and n is a constant (usually set to 1500) to avoid negative numbers when applying logarithms. The resulting high-quality annual optical image data can effectively reduce cloud and noise interference, improving the accuracy of subsequent analysis.

[0085] Step 3.2: After obtaining the median logarithmic ratio of the blue-green band for all images, use the median calculation function to obtain a matrix of median logarithmic ratios for each year's imagery, thus forming a high-quality optical remote sensing image dataset. Actual remote sensing images may contain noise and outliers due to factors such as cloud cover, atmospheric conditions, and sun angle. These noise and outliers can interfere with subsequent water depth inversion and topographic mapping, reducing the accuracy and reliability of the model. Calculating the median matrix effectively filters out these noise and outliers. The median is the value in the middle of a set of data after it is sorted by size. It is insensitive to extreme values and better reflects the central tendency of the data.

[0086] Step 4: Based on the water depth data obtained in step 2 and the optical image dataset in step 3, a quadratic polynomial ratio model (QPRM) is used to construct a water depth inversion model. The water depth data is used as the measured water depth value, and the optical image data is used as the reflectivity data for model construction. Specifically, the following steps are performed:

[0087] Step 4.1: Based on the principle that water has different absorption and scattering characteristics for light of different wavelengths and that blue and green light have relatively good penetration in water, a quadratic polynomial ratio model (QPRM) based on the logarithmic ratio of the blue and green bands is constructed. The quadratic polynomial form can better fit complex actual conditions and can capture more water depth variation characteristics compared to simple linear models, thereby improving the accuracy of water depth inversion. The QPRM formula is expressed as:

[0088] H=a×W md 2 +b×W md +c

[0089] Among them, W md is the median of the logarithmic ratio of the blue and green bands calculated from Sentinel-2 images within a certain time series; H represents the measured water depth data; a, b, and c represent the model fitting coefficients calculated by the least squares method.

[0090] Step 4.2: Use the nearest neighbor method to match the Sentinel-2 and ICESat-2 data based on longitude and latitude. Use random sampling with a ratio of 7:3 to divide the training and validation datasets. Ensure that the data distributions of the training and validation datasets are similar to avoid overfitting or underfitting of the model due to unreasonable data partitioning.

[0091] Step 4.3: Finally, the QPRM model is trained using the training dataset data, and the model fitting results are verified and evaluated using the validation dataset data to obtain a high-precision water depth inversion model.

[0092] Among them, the verification indicators include the correlation coefficient R 2 , root mean square error RMSE and mean absolute error MAE, such as Figure 4 shown.

[0093] Step 5: Use the water depth inversion model trained in step 4 to invert the water depth data of shallow islands and reefs

[0094] Step 5.1: Input the Sentinel-2 data processed in Step 2 into the water depth inversion model trained in Step 4. Combined with the model parameters trained in Step 4, the water depth data at different locations of shallow islands and reefs are calculated to generate an accurate bathymetric map.

[0095] Step 5.2: Subsequently, referencing relevant literature and incorporating a GIS system, the topographic features of the islands and reefs are meticulously annotated using visual interpretation and other methods. During this visual interpretation process, terrain features such as fore-reef slopes, reef flats, lagoons, and beaches are identified based on the islands' geomorphic features, image texture, and color tone. These features are then accurately annotated within the GIS system, completing the topographic mapping of shallow islands and reefs and providing crucial topographic data support for subsequent marine research, resource development, and navigation safety.

[0096] Based on the water depth inversion model established in the above process, the water depth inversion results of active-passive fusion data of Baijiao in 2018 are obtained as follows: Figure 5 The results of Baijiao topographic mapping are shown in Figure 8 As shown; Beizi Island 2018 active-passive fusion data water depth inversion results are as follows Figure 6 As shown in the figure, the topographic survey results of Beizi Island are as follows: Figure 7 shown.

[0097] The above results show that the island and reef shallow-water topography mapping method based on active-passive remote sensing fusion described in this embodiment can meet the needs of high-precision water depth measurement and accurately obtain water depth information of shallow islands and reefs over a large area. In practical applications, there is no need for complex professional equipment or large amounts of manpower, which greatly reduces the cost and difficulty of operation. It can quickly complete the bathymetric mapping tasks of shallow island and reef areas and is highly practical. In particular, it provides important technical support for monitoring shallow-water topography and effectively ensures the efficient and continuous implementation of shallow island and reef bathymetric mapping work.

[0098] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for mapping island and reef shallow sea topography based on active and passive remote sensing, characterized by: include: Step 1: Get active-passive data The active-passive data includes active data and passive data; Step 2: Preprocess the active and passive data separately to extract effective water depth data; Step 2.1: Crop and resample the passive data image, identify clouds, water bodies, and land in the image, and perform cloud and land masking; Step 2.2: Perform DBSCAN photon clustering denoising, refraction correction, and tide correction on the active data to extract effective water depth data; Step 3: Construct an optical image dataset; Step 4: Based on the water depth data obtained in step 2 and the optical image dataset in step 3, a quadratic polynomial ratio model is used to construct a water depth inversion model.

2. The method for mapping island and reef shallow sea topography based on active and passive remote sensing according to claim 1, characterized in that: The active data in step 1 is the Ice, Cloud and Land Elevation Satellite II ICESat-2 dataset; the passive data is the Sentinel-2 dataset.

3. The method for mapping island and reef shallow sea topography based on active and passive remote sensing according to claim 1, characterized in that: The process of cropping and resampling the passive data image, identifying clouds, water bodies, and land in the image, and performing cloud and land masking as described in step 2.1 includes: Step 2.1.1: Crop and resample the satellite imagery data of the downloaded passive data in SNAP; Step 2.1.2: After cropping and resampling the data, use the remote sensing image processing platform to calculate the normalized difference vegetation index. Set thresholds to identify clouds and land. Combined with visual interpretation, create cloud and land masks. Use regions of interest to manually annotate small cloud shadows to form a new mask layer and apply it to the image.

4. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 1, characterized in that: The process of performing DBSCAN photon clustering denoising, refraction correction, and tidal correction on the active data described in step 2.2 includes: Step 2.2.1: Based on the confidence parameter in the photon point cloud file of the active data, remove the photons with a parameter of 0 to complete the coarse denoising; Step 2.2.2: After completing the coarse denoising, use the Gaussian distribution model to perform fine denoising to separate the water surface and bottom photons; Step 2.2.3: Use the density-based clustering algorithm DBSCAN to perform cluster analysis on the photon point group and extract the photon point groups on the sea surface and seabed.

5. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 4, characterized in that: The process of performing DBSCAN photon clustering denoising, refraction correction, and tidal correction on the active data described in step 2.2 also includes: Step 2.2.4: Establishing a data set for the bathymetric mapping task, including: (1) For some DBSCAN algorithm extraction errors, human-computer interaction is used to correct the extraction errors; (2) After obtaining the photon point group data, the improved Parrish water refraction correction geometric model is used to perform refraction correction to obtain the water depth measurement results. The tide correction model and tide data from the tide website are then used to perform tide correction on the bathymetry results and unify them to the mean sea level. (3) Finally, the active data bathymetry results are sorted in ascending order, and one point is extracted every 10 data points as the data set for the subsequent bathymetry and mapping tasks of islands and reefs.

6. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 1, characterized in that: The process of constructing the optical image dataset in step 3 includes: Step 3.1: Construct an image collection of passive data in time series and calculate the median logarithmic ratio of the blue-green bands of all images in each year. The formula is: Among them, R blue and R green are the surface reflectances of the blue and green bands, respectively, and n is a constant; Step 3.2: After obtaining the median of the blue-green band logarithmic ratio of all images, use the median calculation function to obtain the median matrix of the blue-green band logarithmic ratio of each image in each year to form a high-quality optical remote sensing image dataset.

7. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 1, characterized in that: The process of constructing the water depth inversion model in step 4 includes: Step 4.1: Construct a quadratic polynomial ratio model based on the logarithmic ratio of the blue and green bands. The formula is: H=a×W md 2 +b×W md +c Among them, W md is the median of the logarithmic ratio of the blue and green bands calculated from passive data images within a certain time series; H represents the measured water depth data; a, b, and c represent the model fitting coefficients calculated by the least squares method; Step 4.2: Use the nearest neighbor method to match the passive data and active data according to longitude and latitude, and divide the data into training data set and validation data set; Step 4.3: Finally, the quadratic polynomial ratio model is trained using the training dataset data, and the model fitting results are verified and evaluated using the validation dataset data to obtain the water depth inversion model.

8. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 7, characterized in that: When dividing the training data set and the validation data set, random sampling is used according to the ratio of 7:

3.

9. The method for mapping island, reef and shallow sea terrain based on active and passive remote sensing according to claim 1, characterized in that: The method for mapping shallow sea topography of islands and reefs based on active-passive remote sensing also includes: step 5: using the water depth inversion model trained in step 4 to invert the water depth data of shallow sea islands and reefs.

10. The method for mapping island, reef and shallow sea topography based on active and passive remote sensing according to claim 9, characterized in that: The process of inverting the water depth data of shallow islands and reefs in step 5 includes: Step 5.1: Input the passive data processed in step 2 into the water depth inversion model trained in step 4. Combined with the model parameters trained in step 4, the water depth data at different locations of the shallow islands and reefs are calculated to generate an accurate bathymetric map. Step 5.2: Then, combine with the GIS system to mark the terrain features of the islands and reefs.

Images